AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Exploration Is Not Going to Help in This, Um, Transparent New Crimsonaro
Omega the predictor just randomly changing whether you get, like, gum. Put you in a room where boxby is empty, then you probably and then it actually doesn't care what you do in the world where bukby is empty. So it never gets falsified. S what happens with exploration is he the hypothesis, the agent constructs of the environment. It says the following thing: There is one biased coin which is tossed somewhere. And this bias coin is kind of applied to my is kind of sor with the action i intended to take. And alic produces the action that actually end up taking.